What question did this study set out to answer?

The aim is to improve melanoma detection accuracy by using color indices to enhance dermatoscopic images.

April 26, 2026Research on Biomedical Engineering0 citationsOpen Access

A Browning and vegetation index-based deep neural network for melanoma skin cancer classification

EFEdmilson Queiroz dos Santos FilhoUniversidade Federal do Espírito Santo ESEvandro Ottoni Teatini Salles JFJacques FaconUniversidade Federal do Espírito Santo

Key Points

The aim is to improve melanoma detection accuracy by using color indices to enhance dermatoscopic images.
Composite chromatic representations created using three Browning Indices and one Vegetation Index.
Four CNN architectures trained on the Javid Melanoma Dataset consisting of 10,605 images.
Performance assessed using AUC, F1-score, ROC/PR curves, and statistical significance tests.
NASNetMobile achieved an AUC of 96.20% and F1-score of 90.10%.
EfficientNetV2B0 achieved an AUC of 95.50% and F1-score of 89.50%.
Preprocessing with chromatic indices consistently improved performance compared to RGB-only inputs.

Abstract

Cutaneous melanoma accounts for most skin cancer deaths due to its high metastatic potential, making early detection essential. This study investigates whether color indices can enhance dermatoscopic images and improve benign–malignant classification. Three Browning Indices (Aimonino, Fetuga, Lunadei2) and one Vegetation Index (VI) were combined to form composite chromatic representations of RGB images. A Power-Log Based Histogram Equalization technique was applied for contrast enhancement. Four CNN architectures (VGG19, ResNet50, NASNetMobile, EfficientNetV2B0) were trained on the Javid Melanoma Dataset (10,605 images). Performance was assessed using AUC, F1-score, ROC/PR curves, ablation studies, and statistical significance tests. Chromatic-index preprocessing improved performance across all models. NASNetMobile and EfficientNetV2B0 achieved the best results, with AUCs of 96.20% and 95.50% and F1-scores of 90.10% and 89.50%. Ablation studies confirmed consistent gains over RGB-only inputs. The proposed BI/VI-based preprocessing enhances lesion saliency and improves melanoma classification accuracy without requiring segmentation, offering an interpretable strategy for early detection.

AIに質問

Bookmark

View Full Paper